The present disclosure relates to an analysis apparatus, an analysis program, and an analysis system, by which an analysis target object included in an analysis specimen is detected from an image obtained by capturing the analysis specimen.
When observing an analysis target object such as a cell with a microscope, for example, if the cell is floating in culture, an observer needs to search for it. However, a field of view is small at a high magnification of the microscope, and hence the observer needs to search for the analysis target object while largely moving a field-of-view range. Meanwhile, if the size of the analysis target object is small at a low magnification of the microscope and the analysis target object is lost in air bubbles and the like, and hence the search needs high personal costs.
For example, even if time-lapse imaging in which the analysis target object is imaged over time is performed, there is a low possibility that the analysis target object remains at the same position during capturing images. Therefore, the observer needs to search for the analysis target object every time when an image is captured. In addition, there is a fear that the analysis target object searched out by the observer is different from the previously imaged object.
In view of the above-mentioned problems, techniques of automatically searching for an analysis target object have been developed. For example, Japanese Patent Application Laid-open No. 2010-169823 (hereinafter, referred to as Patent Document 1) describes a technique of predicting a motion of a previously found cell or the like by a motion prediction model and utilizing it in cell tracking.